The broad objectives of the proposed research are to assess the power of psychological antecedents to predict aggressive behaviors in middle childhood and to determine how specific variables act as mediators between environmental factors and subsequent aggression. The objectives are related to the long-term goal of generating more effective interventions to lesson adolescent violence in ethnically diverse populations. The main hypotheses are: (I) certain key environmental variables and a history of aggression in children predict aggressive behaviors and act as a baseline for the following hypotheses. (2) Beyond the predictability of environmental variables and aggressive history, a negative core identity in relationships predicts aggressive behaviors. (3) Shyness in boys predicts aggressive behaviors when combined with a negative core identity in relationships and aggressive backgrounds. The same pattern in girls predicts a tendency toward victimization. (4) Beyond the predictability of environmental variables and aggressive history, lower imaginativity in children predicts aggressive behaviors, especially when combined with a negative core identity in relationships and an outgoing temperament. To test these hypotheses, 582 children between 7 and 13 years of age and their mothers will be interviewed to assess specific predictor and outcome variables. Children will be tested from lower to upper SES levels in a representative sample taken from African-American, Latino, and white urban families. A second assessment will be completed at a one-year follow-up.